|
How can a covariance matrix C be used with a new test sample 'x' to make a prediction? A friend of mine was trying to show me the following: (x-u)' C (x-u) Any idea what he was talking about? |
|
This is the (squared) mahalanobis distance between x and u. exp(-this) is proportional to the gaussian probability of x. I don't understand what this has to do with prediction, however (unless in the case of fisher discriminants, in which you choose the class that minimizes that). Wasn't your friend talking about gaussian processes? They use covariance matrices (actually covariance functions, sampled appropriately in a matrix), but in a rather different way. |
|
no, it must have been Mahalanobis distance. Thank you! What he was doing was he created a Covariance matrix for each of the class distributions, then used this measure to classify a new test point. 1
That is the fisher discriminant: http://en.wikipedia.org/wiki/Linear_discriminant_analysis
(Sep 12 '10 at 00:28)
Alexandre Passos ♦
|
